Accessibility cross audit: why have a score checked again?
The short answer
A cross audit is an independent counter-audit: a second auditor checks an accessibility score that's already been announced. It's useful after fixes made in-house, or to confirm another provider's result. It can reveal a lower score than expected, for example because the criteria were misinterpreted.
What is a cross audit?
It's a second audit, carried out by an auditor independent of the one who did the first, on a website that's already been assessed. The aim isn't to redo the work for its own sake: it's to verify that the announced score and conformity status hold up.
On the official side, DINUM lists cross audits among the ways of ensuring an audit is reliable. It doesn't require them or set a method for them.
When is it useful?
In my practice, two situations come up.
Your teams have fixed the website. An audit listed the non-conformities, your developers dealt with them, and now the score needs revising. An outside view checks that the fixes really resolve the criterion, not just the symptom.
You want to confirm a provider's score. You have a figure and you're about to commit to it: publish it, report it to management, put it in a tender. You want to be sure it holds before you do.
If you simply want to know where your website stands, with no audit to check, that's more the role of the Flash Audit.
What does a cross audit reveal?
I've found scores lower than the initial audit, including after a round of fixes. The cause was nothing dramatic: a misinterpretation of the criteria, by the client or by the initial provider.
That risk makes sense, because the calculation is strict. A criterion is only validated if it passes on every page of the sample, as I explain in how the conformity rate is calculated. A few misunderstood criteria can therefore shift the score by several points.
A lower score isn't bad news in itself. It's the real starting point, and it stops you fixing the wrong things.
Why check before publishing?
Because your accessibility statement relies on that score. Under the RGAA, it states whether the website is fully, partially or non-compliant, and the line between "partially" and "non-compliant" is set at 50% of the criteria. If a score is overestimated around that line, the published statement becomes inaccurate.
What do you receive at the end?
Two deliverables, built from the existing documents:
- the updated audit report, with the counter-audit's observations;
- the updated accessibility statement, with the verified score.
A cross audit starts from the initial audit's document. So you keep a single reference document, and you can see exactly what has changed. The audit process stays the same: sample, validation, evaluation, delivery.
The options are on the Accessibility page.
Frequently asked questions
Is a cross audit mandatory?
No. DINUM, the French digital directorate, mentions it as one way of ensuring an audit is reliable, alongside using an external provider or training internal experts. No text I've consulted requires it.
What's the difference from a full audit?
A full audit starts from scratch and produces the compliance plan. A cross audit starts from an existing audit and checks its result. So it doesn't replace the plan of fixes, but it tells you whether the score you're relying on is right.
Can a cross audit lower my score?
Yes, if the initial score was overestimated, and I've seen it happen, even after a round of fixes. It's better to find out at that point than after publishing an inaccurate accessibility statement.